European Races

[Cambrius (The Red)]

Yes, facts ARE facts.
Iberians are 'Meds.', while Greeks are 'Alpines' in phenotype.

And in general Nordics and Alpines are closer to each other than Meds. to Nordics. Just take a look on Eastern Europe!

Sure, dream on. Greeks are among the most Mediterranean of population groups in Europe. Phenotypically they run the Balkanic / East Med spectrum: from Dinaric to Levantine.

It's been abundantly demonstrated through numerous DNA projects that Greeks are far more Med / Southern Euro (autosomal DNA governs phenotypic structures) than Iberians, who are majority Atlantic. That's a fact, nothing bad or good about it. If you can't accept the reality of the situation it's your loss. Time to move on.

 

[sparkey]

Well ok, but it's impossible to deny it means European (call it Med. or another name, ¿who cares?). I did not create the distances as you can imagine, and the examples I mentioned comparing West Euro and Mediterranean are perfectly clear.

Oh, I don't think that just calling it "European" is disingenuous. That seems to be an appropriate geographic descriptor.

 

[Knovas]

Oh, I don't think that just calling it "European" is disingenuous. That seems to be an appropriate geographic descriptor.

I'm not telling you must call it plain European, I'm telling it represents Europe, more likely Southern Europe since it's where it peaks. I don't care about the name, it's just part of Europe, the same as the other two.

Results are proportional to the other runs. If you check Northern Euro's you'll see some of them also increased the average, others are the same, and others decreased. But if you are curious, you can go and check that some Baltics and Polish are the ones who increased more with the East European...so it's not that "easy" to score high Mediterranean as some people seems to think.

However, we can discuss if you want which component is more European taking the distances and comparing with other analysis. For example:

DOD725 is 54.4% Med, 38.2% West Euro, and 4.1% East Euro in the last run. At K=10 he was 37.4% North Euro and 52.3% South Euro. Let's see then...

East + West = 42.3% - 37.4% (N. Euro) = 4.9%

Medit. = 54.4% - 52.3% (S. Euro) = 2.1%

Another one, DOD217, had 36.4% Northern Europe and 50.6% Southern Europe. K=12 show 46.4% West Euro (+10%!!!), and 47% Med. Note this one has even decreased the Mediterranean score.

Well, curiously, the East and West Euro (Northern Europe) increased more than the Mediterranean (Southern Europe) compared with K=10. Both are Iberians, a "close" reference for Basques.

Time to think about it guys.

 

[Dorianfinder]

West-Asian peaks in the region of Caucasus/Anatolia/Middl-east. The mediterranean is strong in ALL of Europe, for example Orcadians have 25%, Scandinavians 15-20%, etc and peaks in Sardinians, North-Italians. The basques who are the purest Europeans have also 45% mediterranean. West-Asian is small in all of Europe, except those countries with lot of near-east influence like Greece or Italy

Keep in mind that a high inbreeding coefficient combined with a high genetic fitness should for all intents and purposes give a 'false' homogeneity as foreign elements are diffused within a contained population sample; this is one of the reasons why Basque and Sardinian are used as admixture groups on Dodecad, not because they are 'pure' but simply because they are 'typical' admixture groups of their respective localities. Coupled with this we also need to bear in mind that Basque is an admixture group used to illustrate the west Mediterranean component in various populations.

I mean "the Basques are the purest Europeans" on what level?

Vernesi (2004) compared mt-dna haplotype frequencies and found that Etruscan mt-dna has an admixture of those mt-dna haplotypes found in the Turkish, Basque and North African populations. This is merely one of many mt-dna studies that suggest a high degree of admixture in these communities.

The majority of Basque admixture in the Dodecad project is not visible because of the calibrations used within the program set to incorporate major genetic input and incorporate it into the Basque factor-analysis. All genetic information typed to determine the typical Basque admixture is indicated as 'Basque' with only 'very' unique genetic input visible, these depend on the settings and regions compared.

It is important to note that each Dodecad study compares different regions, requiring a different calibration and admixture setup. For example when Mediterranean countries are compared with Anatolia and ME countries we find the Basque population displays little North European, Dagestan, West Asian and Sardinian. However, in a different Dodecad analysis comparing Basque with Eastern European countries, we find a very high level of North European in the Basque population as North European does not include the Northeast European populations under comparison. Therefore North European in this example would entail other regions that display a high Alpino-Celtic admixture, as a consequence reflecting a higher level of 'North European' than would otherwise be seen.

In conclusion, Dodecad utilizes complex algorithms and makes assumptions based on its programmed calibrations. The data is visually represented and appears easy to comprehend, however, various statistical issues make analysis tricky, especially when comparing similar populations. There is however no doubt in my mind that as time passes the Dodecad program will become better defined in its interpretation and visualization of same-type admixtures.

 

[Wilhelm]

Is not just at Dodecad, also at Eurogenes and at academic studies.

 

[Drac]

In general I do consider the Greeks much more 'Nordic' than the Iberians. And I'm not looking at their skin-color, but to their skeleton (and cranium) and how big they are. Greeks have more a stature/shape of the warriors than the Iberians. Iberians are more 'fragile'.

Actually it's the other way around: Mediterraneans are so close to Nordics cranially that anthropologists have a very hard time telling them apart. Both of them are dolichocephalic, while Alpines are brachycephalic, very easily distinguished from Nordics and Mediterraneans. This strong similarity in skull shape convinced Carleton Coon that Nordics were really nothing more than taller and more depigmented Mediterraneans.

The average height among Greeks is the same as Spaniards: 5 ft 9 in - 5 ft 10 in

Greeks, by the way, are not wholly Alpine by any stretch of the imagination. They have strong Mediterranean presence as well.

 

[Knovas]

Is not just at Dodecad, also at Eurogenes and at academic studies.

And 23andme also, as I said, show the Basques as the population with less African and Asian similarities. It's fact this genetic is the most purest European form we can find, all indicators follow that way. Even DOD818 with a little admixture (96% Euro), is less similar than Baltics or any other European to Africa and Asia, just because the huge Basque percent deviates him far from those groups. No need to say a pure Basque, or just one with 99% European (average) would show incredibly low numbers while checking African and Asian groups.

Checking Dodecad K=12 v3 it's easy to notice they have a "perfect" equilibrium in average, showing more than 52% West European (far from Africa) and more than 45% Mediterranean (far from Asia and not precisely close to Africa, specially Negroid groups). More or less, it matches the cluster they get, althought the best way to fit this population is using primarly a Basque cluster. I also think Iberians need a Basque cluster to get accurate results, since they have substantial Basque ancestry, specially in the Northeast side. The same is valid for the Southern French.

 

[Dorianfinder]

Is not just at Dodecad, also at Eurogenes and at academic studies.

The principles and methodologies have similar issues. Whether Autosomal tests focusing on genetic percentages as determined by examining SNPs or standard identity markers, such as the CODIS profile in combination with databases such as OmniPop, ENFSI and adaptations of various published studies and commercial testing agencies. The Maths for the most part is linear and seeks to analyze commonalities and express them as percentages. Much room exists for misrepresentation of certain standard identity markers under the CODIS system. For example, not too long ago R1b was touted as a palaeolithic European marker, now we understand that such conclusions were the result of poor definition of R1b subclade distributions and a general lack of non-European samples. The same is true for standard identity markers being used to identify Autosomal admixtures in various populations.

 

[Knovas]

If you were right, most Iberians and Southern French would cluster with Basques since they seem to have appearently very similar composition (just see k=12 v3 run). However, things doesn't work like this, because most of them are only PART Basque (and usually quite ancient). Basques have a very specific genetic, that's why sometimes Davidski (Eurogenes) rejected to include them in the runs: just to avoid spurious results.

In my opinion this K=12 is not valid for the Basques, the best run for them MUST include a Basque cluster (like the other K=12 did)...I know perhaps it's boring, but it's what really fits them better.

 

[Wilhelm]

The principles and methodologies have similar issues. Whether Autosomal tests focusing on genetic percentages as determined by examining SNPs or standard identity markers, such as the CODIS profile in combination with databases such as OmniPop, ENFSI and adaptations of various published studies and commercial testing agencies. The Maths for the most part is linear and seeks to analyze commonalities and express them as percentages. Much room exists for misrepresentation of certain standard identity markers under the CODIS system. For example, not too long ago R1b was touted as a palaeolithic European marker, now we understand that such conclusions were the result of poor definition of R1b subclade distributions and a general lack of non-European samples. The same is true for standard identity markers being used to identify Autosomal admixtures in various populations.

You are wrong, it has nothing to do. The components are not decided by humans, they are made automatically by software based on commonality of SNP's in the populations. Then the humans just labels these components, of course the label is arbitrary, but sometimes is very clear, like when it peaks at more that 70% in a certain population.

 

[Knovas]

Yes, clusters come without names. And Basques are simply Basques, not casual or forced they're always listed in a different cluster, althought it's true some of them have a little of other ancestries. For example, I don't expect DOD818 to be less than 90% Basque (enough to deviate in comparison with typical Europeans), even knowing he's slightly mixed.

I said this a lot of times: too bad Dienekes' is not running the other analysis. For some people (not only Basques), that was more accurate to define the European ancestry. It's true other clusters behaved a bit strange, but nothing serious.

 

[Dorianfinder]

You are wrong, it has nothing to do. The components are not decided by humans, they are made automatically by software based on commonality of SNP's in the populations. Then the humans just labels these components, of course the label is arbitrary, but sometimes is very clear, like when it peaks at more that 70% in a certain population.

When an SNP is programmed into the system it does not differentiate beyond known subclades. This is known as poor definition and results in numerous overlaps between various regions. If I use a common denominator to anchor my sub-groups it means this common denominator or 'standard group' will show up on the factor analysis as being between 0.95 and 1.0 ... use caution not to interpret this as a percentage of admixture regarding the 'standard group' because it isn't representative as it only represents and includes commonalities within the chosen population.

For a study to be valid it needs to test what it claims and any influence from confounding variables should be insignificant. Dodecad and Eurogenes both display serious validity issues as can be observed when similar populations are tested. Similar populations need a high resolution technique to differentiate their unique Autosomal structures, this is still some way off and what we have now is still very elementary, but fun to look at and discuss on forums. Those who say that a cluster comes out with no name are indeed right in saying so, however they are failing to mention that the factors or 'measurements' used to formulate each cluster is determined through tests that determine certain marker values, very few I may add, along a poorly defined allele.

An example of how the Dodecad program works is when you have 10% R1a in two populations. The one population, lets call it A, has a high proportion of the SNP M458, while B has a much smaller proportion of this SNP M458. Manually, we can calculate the average and estimate it using a simple formula. Dodecad does this using Autosomal values. It is not an exact science as base samples used to calculate norms are fickle and prone to sample error.

An anecdotal example could be used to illustrate what poor definition entails. When there is a man who is believed to be the world's heavyweight boxing champion. Many people may in fact be better than him but because they do not have an opportunity to challenge him he will be proclaimed world champion. He is in fact only the champion of a small sample of boxers. Similarly, it is often not the case that this that or the other group display less admixture than other populations rather than neighboring populations who have not as yet been Autosomaly mapped to a similar extent. Iberia has a very rich genetic heritage and is no less effected by surrounding populations than Greece. The difference comes in that whereas Greece's neighbors share high levels of Anatolian genetic code the Iberian population has not distributed its genetic heritage to the same extent, making it 'purer'. One could argue that Greece's success at colonizing and distributing its genetic heritage throughout the region now appears on admixture analyzes as though it had in fact been conquered and added to, when the exact opposite may in fact be the more likely scenario.

 

[LeBrok]

Very interesting Dorianfinder, and it plainly illustrates shortcomings of this and similar analysis, good for blogging but not exact science.
One other thing that bothers me is the starting point for European ancestry. What is a definition, all could agree, to be called European?
Can we agree that today's European is everyone with european citizenship?
What about recent huge wave of emigrants from Middle East and Africa, getting ot 10 or 20% level in some areas in Europe? After all if they have citizenship it means that they were accepted to become Europeans buy people or their authorities.
If starting point is that citizens are europeans, therefore all citizens need to be included into statistical studies. This will definitely change genetic map of Europe, and autosomal relation of any country to continental admixtures.

If all citizens shouldn't be included,then how long newcomers and their decedents have to live in Europe to become Europeans and be included in dna statistics? Hundred years? I'm having doubts that even a thousand years are not long enough of our Iberian friends. After all they have a problem including Gipsys into equation. Gipsys would definitely pull Iberia south, a big no no.

So, where do we start to find a pure European for study comparison.
Bronze Age, Iron Age? But wait, millions of impure IE from Asia invaded the continent. Too many fresh emigrants.

Paleolithic? So only I is true europen. Should we include R1a if we find it at that time on west side of Urals?

Should we start with CroMagnon? But wait! Neanderthal was the purest European, after all they lived in Europe for almost a million years. The person with most Neanderthal autosomal dna should be the purest european, if you care.

 

[Knovas]

You are talking about ancient times when this genetic components surely did not exist, but the main point is that it's plain impossible to put those populations at the same level of the modern ones for obvious reasons. I could say following your logic that some of the "homos" found in Atapuerca who lived 800.000 years ago are the purest Europeans...fantastic argument! LOL. Also, Neanderthals were present in the Near East and very deep in Asia too, so not only "Europeans", the thing is a bit more cosmopolitan as you can imagine.

And another thing, is that 23andme works with Genome Wide comparisons. Then, the margin of error is quite low and the pattern is the same for all people. Greeks show less European similarity because they are more mixed than for example French or Iberians, doesn't matter the antiquity of the admixture and if Greeks colonized Anatolia (with the consequent and logical genome coincidence at some points). That's fact, they carry substantial Near Eastern, but we can discuss how is this possible if you want.

Basques at 23andme show the highest similarities considering BOTH Northern and Southern Europe, and they are the Europeans with less non European affinities. Remember we are refering to Genome Wide coincidence here, so I'm afraid what you try to argue is simply wrong.

 

[Dorianfinder]

Remember we are refering to Genome Wide coincidence here, so I'm afraid what you try to argue is simply wrong.

Remember that 23andMe does NOT sequence the entire genome, it uses genotyping instead, which is an SNP-based technique. Trust me when I say that I understand your resistence to the thought that statisticians are responsible instead of a super-computer, it's always easier to believe results are machine-based rather than subjective and prone to human fallibility.

A recent study presented at the European Society of Human Genetics concluded that both companies (23andMe & DeCode Genetics) sell inaccurate predictions of disease risks to their customers*.

*Direct-To-Consumer Genetic Tests Neither Accurate in Their Predictions nor Beneficial to Individuals, Study Suggests. 31st May 2011. http://www.sciencedaily.com/releases/2011/05/110530190344.htm

Below is an admission of the limitations of 23andMe's ancestry division by the company co-founder Linda Avey. She goes on to explain some important points about how the links are established, who establishes them, and confirms some of the issues they experience, namely representation, definition and interpretation of results.

Linda Avey, co-founder of 23andMe wrote:

... it's just the interpretation of what the datapoints means which is, in some cases, early and subject to revision.

Getting wider diversity in genetics research is a key objective of 23andMe...need to figure out waysto attract people of all backgrounds. Once we have enough representation of any given population, the statisticians can see if genetic associations replicate (or discover new ones!).

 

[Knovas]

Probably some changes would come, of course. However, I don't expect an incredibly great discovery refuting what it's now clear about the Basques. At least, if some interpretation changes focussing on them, it's very likely other populations can be affected for the same reason. And we can go on again and again with this...

Disease Risk goes apart from ancestry, and I don't think results are inaccurate at all, the only "problem" is that it's impossible to assure you'll suffer a concrete illness (I supose that's what they call inaccurate, ¿what else?)...but that's something all people knows or, at least, should know. I can say they listed some Disease Risks wich are present in my family, but you are free to think what you want.

Also, trying to discredit 23andme, clustering studies, admixture analysis, etc., etc...you know, it doesn't look too much serious suggesting all people is wrong.

 

[julia90]

Externally i have a very mediterranean phenotype, dark brown hair and brown eyes, as well as having a convex nose.

my sking isn't olive, it's quite pale.

my profile as well as hair colour is quite similar to this famous italian girl (on the right):

also frontally my face is similar too:

i consider my self akin to dinaromorphic-mediterranean people from italy and the balkans.. i would be typical there.

In my family i have a strong, if not dominant prevalence of this kind of genes.
Mediterranean-dinarid
alpinoid
as well as CM or berid tendencies on some members, like my father and my brother

my father was classified as Adriatid with CM tendencies, my brother as med-berid by people on other sites.
my mother as a kombo like me.

about hair colours, in my family all recorded people were dark brown, with only one grand mother with light brown-blondish hair.
Eye colours change much more, with half of my great-grandparents having light eyes

 

[Knovas]

First Mediterranean listed in the thread. Welcome

 

[Mzungu mchagga]

I'm a depigmented Mediterranean.

 

[julia90]

thanks, althought my head index should be mesochephalic.
my father's head is typically dinarid, high but flat behind.
my brother is very dolichocephalic
my mother is mesochephalic, but more on the brachimorphic side.